feat: add MiniMax as first-class LLM provider

[octo-patch]

Summary

Adds MiniMax AI as a first-class LLM provider for OntoGPT, accessible via their OpenAI-compatible API.

What's changed

• LLMClient (src/ontogpt/clients/llm_client.py): Detect MiniMax provider via minimax/ model prefix or --model-provider minimax, auto-configure api_base to https://api.minimax.io/v1, resolve API key from MINIMAX_API_KEY env var with oaklib minimax-key fallback, clamp temperature to MiniMax's required (0.0, 1.0] range, and route through litellm's OpenAI-compatible path
• Model registry (src/ontogpt/__init__.py): Register MiniMax-M2.7 and MiniMax-M2.7-highspeed models (204K context window) in the model cost map
• CLI (src/ontogpt/cli.py): Update --model-provider help text to mention MiniMax
• README: Add MiniMax setup and usage documentation section

Usage

# Set API key
export MINIMAX_API_KEY="your-key"

# Use with minimax/ prefix
ontogpt extract -t drug -i example.txt -m minimax/MiniMax-M2.7

# Or with --model-provider
ontogpt extract -t drug -i example.txt -m MiniMax-M2.7 --model-provider minimax

Test plan

• 25 unit tests covering provider detection, API key resolution, temperature clamping, completion parameter forwarding, system message handling, and model registry
• 3 integration tests verifying real API calls with MiniMax-M2.7 and MiniMax-M2.7-highspeed (skipped when MINIMAX_API_KEY is not set)
• All 11 existing test_llmclient.py tests still pass
• All 28 new tests pass